In the United States an estimated 70 million amalgam restorations are replaced annually. Between 30 and 36 percent of these replacements are due to recurrent caries. Progressive fracture of the margins of amalgam restorations is thought to promote microleakage and to lead to recurrent caries. The goal of this research program is to determine the causes of marginal breakdown. Amalgam restorations have been placed in dentrue teeth mounted in free-end saddle removable partial dentures. This has enabled the investigators to retrieve sets of restorations at frequent intervals. By examining cross sections beneath fractured margins, microstructural changes that lead to breakdown have been followed. Two interesting findings have resulted: 1) marginal fracture is accelerated in the denture tooth model, occurring before significant corrosion or phase transformations have taken place and 2) at least some of the cracks which lead to fractured margins nucleate at low strain rates. In the this new research program, the investigators will continue to study two representative high copper amalgams. The denture tooth model will be used to study marginal fracture at proximal margins of Class II restorations and at the margins of burnished and unburnished restorations. Although over 90 percent of recurrent caries is associated with proximal margins, little is known about breakdown at these margins. The denture tooth model permits impressions of proximal margins to be made. Patients will be recalled for replication or retrieval of restorations at 1, 4, 12, 24, 36, 48, 72, and 96 weeks. Epoxy replicas made from successive impressions will enable the investigators to compare the rate of breakdown at occlusal and proximal margins. Restorations in denture teeth will be compared with restorations in retrieved natural teeth which have been in service more than five years. This study will determine how breakdown differs when corrosion and phase transformations are and are not playing significant roles. In each of these retrieval studies, the subsurface microstructure will be characterized by scanning electron microscopy and x-ray energy dispersive analysis, Data on the modes and sites of crack nucleation will be sought. To learn more about crack nucleation at low strain rates, a variety of types of amalgams will be tested at strain rates between 1x10-6 and 1x10-5 min-1. Microstructural change during testing will be followed using scanning electron microscope. The goal will be to determine how crack nucleation varies as a function of microstructure and strain rate.